Scott A. Golder - Extended abstract of a paper to be presented at AoIR 2004.

The full paper is available here.

Communities are complex social structures that change and grow over time, as a result of the behavior of their constituent members. This behavior is not random, however. Each individual has a set of social skills, privileges and responsibilities within the community that describes his or her behavior. This behavioral repertoire is called a role. In this paper, we explore the idea of a role as an analytical tool for studying communities in Usenet newsgroups. We develop a taxonomy of roles based on social and linguistic behavior, and use this taxonomy to describe how changes in the social structures of newsgroups, including group norms and hierarchies, take place.

For several weeks we conducted an ethnographic study of approximately sixteen newsgroups as unobserved observers. The newsgroups were chosen arbitrarily, with the proviso that the groups be unmoderated, have a significant volume of conversation, and represent a wide variety of topics. These topics included technical fields like chemistry and computer programming, hobbies like gambling and music, and social issues like parenting and religion. During the observation period, we compiled data about each participant and about numerous conversation threads. These data formed the basis for a taxonomy of roles, which we used to analyze social interactions and their impacts on their respective communities.

In everyday life, there are two kinds of roles, those that are formal and explicit, and those that are informal and largely unspoken. The former category includes unambiguous, societally-defined roles like President and police officer, as well as family- and gender-based roles like parent, husband and daughter. The latter category includes roles that are less rigidly defined. In a group of friends, one might have the unspoken role of planner, always coordinating outings and making reservations. Such a role develops not through assignment, but rather through repeated social interaction. Predictable patterns are implicitly and mutually agreed upon, and become social constraints, however flexible, that frame our interactions in everyday life. In any social interaction, an understanding of the participants' roles is therefore essential for contextualizing and understanding their behavior.

The roles people assume in Usenet are interesting precisely because they fall into the second category; except for moderators, all roles people play are implicitly and continually negotiated within the community. A role is reflexive, both determining and being determined by one's linguistic performance. For example, one who continually provides assistance to others may rise to become a leader in the community and, as a leader, one may have the privilege of speaking on behalf of the community. If someone who did not enjoy that privilege believed they did and attempted to speak authoritatively, others may voice their disapproval, rejecting the individual's claim to a role other than his or her own. Such interactions - those in which conflict occurs - are useful because they make unspoken social rules more salient. Despite being interesting for researchers, these conflicts have real effects on the community members, as they struggle to negotiate social positions comfortable for themselves and for the rest of the community.

In developing our taxonomy, we identified several roles and several measures of behavior that describe them. Some of the roles we identified are commonly recognized by Usenet communities; others are categories of our own devising. What we see as vital is not this particular set of roles, but rather the recognition that participants' behavioral repertoires are useful tools for understanding and contextualizing interaction within the newsgroup community. The roles we describe include Newbies, Celebrities, Elders, Trolls and Ranters. We discuss each of these roles with respect to the behavioral traits they tend to employ.

The behavioral measures we have identified come primarily from sociolinguistics and social psychology. First, we consider Hymes' notion of communicative competence [1], or mastery of the style of speech in a particular speech community. More than simply being able to form a syntactically correct or understandable sentence, communicative competence requires understanding the community's beliefs and attitudes, and the ability to present one's thoughts in a manner consistent with those beliefs and attitudes. Secondly, we consider the strategies an individual might choose in his or her interaction, of which Jones and Pittman [2] propose five: ingratiation, intimidation, supplication, self-promotion and exemplification. Lastly, we consider the effects (and causes) of participation inequality [3]. Some Usenet participants post tens or hundreds of messages a day, while others post only once and leave. The sheer volume of a user's participation affects how much of an impact the user has on the rest of the community. It is virtually impossible to participate in a newsgroup in a meaningful way without encountering the posts of the most prolific members.

The effects of participants' roles are most visible when considering the creation and manipulation of social boundaries in the community. These boundaries dictate who is part of the community and what is an acceptable topic. They are defined through the behavior of the participants, and are made clear most especially when one such boundary is transgressed. We note that participants with higher social status are most able to defend those boundaries, but are also most able to push those boundaries, as well. It is through the repeated negotiation of such boundaries that communities change over time. We develop this idea by presenting several case studies.

We propose that a role-based approach to understanding communities and their participants yields an account that better explains the changes that take place, by allowing for generalization of behavior across dissimilar communities. In an environment like Usenet, where any social structure must arise solely through the actions of its members, understanding the roles that the members develop is essential in understanding the resulting community.

1. Hymes, D. Foundations in Sociolinguistics: An Ethnographic Approach. University of Pennsylvania Press. 1974.

2. Jones, E. and Pittman, T. Toward a General Theory of Strategic Self-Presentation. Psychological Perspectives on the Self. Ed. J. Suls. Erlbaum, 1982.

3. Whittaker, S., Terveen, L., Hill, W., Cherny, L. The Dynamics of Mass Interaction. Proc. CSCW 1998.